As data delivery configurations via communication networks, configurations referred to as “P2P” have been recently widespread, In conventional content deliveries, users download files from servers managed by their carriers or the like. However, in P2P content distributions, no servers that collectively store therein digital contents are provided, and users can exchange their digital contents stored their personal computers therebetween.
FIG. 1 illustrates a P2P network 90 that includes some nodes (computers), ranked as upstream nodes, connected to high communication networks and alternative some nodes (computers), ranked as downstream nodes, connected to low communication networks. The P2P network 90 is configured such that contents and their index information are more gathered in the upstream nodes to thereby enhance the content-search efficiency or the content-delivery efficiency (FIG. 1 is a schematic view, and many layered networks may be provided downstream of the downstream nodes or upstream of the upstream nodes, or a two-layered network of upstream and downstream may be provided).
In the P2P network 90 illustrated in FIG. 1, when a node for retrieving a content gives a query command to an upstream node adjacent thereto, the upstream node is adapted to search its managed content list for the requested content; this content list includes collected index information on contents). The upstream node is adapted to give, to downstream nodes, the locations of nodes, as an answer; these nodes disclose the requested content. When the requested content is not included in its managed content list, the upstream node gives the query command to a higher upstream node or a coordinate upstream node.
When receiving a result of the search from the upstream node, the downstream node connects to a node described in the answer to download the content.
For example, P2P file sharing software of Winny that forms the P2P network 90 generates tag information associated with contents in order to distribute the contents; this tag information for a content describes the location of a node storing the content (IP address, port number; hereinafter referred to as file location information). The Winny distributes the tag information to upstream nodes adjacent thereto.
When receiving the tag information, an upstream node distributes the tag information to a higher upstream node or a coordinate upstream node. This allows many pieces of tag information to be collected upstream, so that upstream nodes are adapted to manage the tag information lists, that is, the content lists.
In the P2P network 90 formed by the Winny, many pieces of tag information are more stored in the upstream nodes.
In addition, in the Winny, during the distribution of tag information, the tag information (file location information) is updated with a fixed probability. For this reason, when receiving the updated tag information and trying to download a content, a node provides a download request to an upstream node described in the updated tag information. However, because the upstream node does not actually store the content, the upstream node searches its content list, downloads the content from a node that actually stores the content, and uploads the content to the download-request providing node.
This results in that the queried content is downloaded to the source node of query, and the content is also stored in the upstream node that relays the distribution of the content. In the P2P network 90 formed by the Winny, upstream nodes serve as relay nodes of content distributions, and, therefore, store many published contents (nodes that have downloaded contents serve as content-publishing nodes).
These will be specifically described with reference to FIGS. 1 and 2.
FIG. 1 illustrates a situation in which a node S has opened a content α0, and a node B tries to download the content α0. As illustrated in FIG. 1, tag information β0 on the content α0 indicative of the node S holding the content α0 is distributed to upstream nodes located more upstream relative to the node S.
First, the node B transmits a query for the content α0 to an upstream node D adjacent thereto. When receiving the query for the content α0, the node D returns the tag information β0 to the node B because it holds, in its content list, the tag information β0 indicative of the location of the content α0.
This results in that the node B finds the location of the node S that has published the content α0 from the received tag information β0, directly establishes a P2P connection to the node S, and downloads the content α0.
Moreover, FIG. 2 illustrates a situation in which a node S has opened content α0, and a node C tries to download the contents α0. As illustrated in FIG. 2, the tag information β0 on the content α0 indicative of the node S holding the content α0 is distributed to upstream nodes located more upstream relative to the node S. In addition, the tag information β0 is updated in an upstream node so that the tag information β0 is changed to tag information β0a indicative of a node A holding the content α0.
First, the node C transmits a query for the content α0 to an upstream node E adjacent thereto. When receiving the query for the content α0, the node E queries the node A located adjacently upstream relative thereto because it holds, in its content list, no tag information indicative of the location of the content α0.
Because holding, in its content list, the tag information β0a indicative of the location of the content α0, the content A returns the tag information β0 to the node C via the node E.
This results in that the node C finds the location of the node A that has published the content α0 from the received tag information β0a, directly establishes a P2P connection to the node S, and requests it to download the content α0. However, because the node A does not actually hold the content α0, it downloads the content α0 from the node S that actually holds the content α0, and uploads the content α0 to the node C. As a result, the content α0 is transferred to be located in both the node C and the node A, making it possible to improve the efficiency of distributions of contents.
First patent document: Japanese Patent Laid-Open No. 2006-178782
Non-patent document: The technology of Winny, Isamu Kaneko, ASCII, October, 2005